💄 improve string output of documents

Author: ianardee

mindee/__main__.py

@@ -170,15 +170,16 @@ def _parse_args() -> Namespace:
             "--no-cut-pdf",
             dest="cut_pdf",
             action="store_false",
-            help="Don't cut the PDF",
+            help="Don't cut document pages",
         )
         subp.add_argument(
             "-p",
             "--pdf-pages",
             dest="pdf_pages",
             type=int,
             default=3,
-            help="Number of PDF pages to cut by, default: 3",
+            choices=[1, 2, 3],
+            help="Number of document pages to cut by, default: 3",
         )
         subp.add_argument(dest="path", help="Full path to the file")
 

mindee/documents/financial_document.py

@@ -188,8 +188,8 @@ def __taxes_match_total_incl(self) -> bool:
             return False
 
         # Reconstruct total_incl from taxes
-        total_vat = 0
-        reconstructed_total = 0
+        total_vat = 0.0
+        reconstructed_total = 0.0
         for tax in self.taxes:
             if tax.rate is not None and tax.rate != 0 and tax.value is not None:
                 total_vat += tax.value

mindee/documents/invoice.py

@@ -124,8 +124,10 @@ def __str__(self) -> str:
         customer_company_registration = "; ".join(
             [str(n.value) for n in self.customer_company_registration]
         )
-        payments = ", ".join([str(p) for p in self.payment_details])
-        taxes = ", ".join(f"{t}" for t in self.taxes)
+        payment_details = "\n                 ".join(
+            [str(p) for p in self.payment_details]
+        )
+        taxes = "\n       ".join(f"{t}" for t in self.taxes)
         return (
             "-----Invoice data-----\n"
             f"Filename: {self.filename}\n"
@@ -139,7 +141,7 @@ def __str__(self) -> str:
             f"Customer name: {self.customer_name}\n"
             f"Customer company registration: {customer_company_registration}\n"
             f"Customer address: {self.customer_address}\n"
-            f"Payment details: {payments}\n"
+            f"Payment details: {payment_details}\n"
             f"Company numbers: {company_numbers}\n"
             f"Taxes: {taxes}\n"
             f"Total taxes: {self.total_tax}\n"
@@ -174,8 +176,8 @@ def __taxes_match_total_incl(self) -> bool:
             return False
 
         # Reconstruct total_incl from taxes
-        total_vat = 0
-        reconstructed_total = 0
+        total_vat = 0.0
+        reconstructed_total = 0.0
         for tax in self.taxes:
             if tax.value is None or tax.rate is None or tax.rate == 0:
                 return False
@@ -211,8 +213,8 @@ def __taxes_match_total_excl(self) -> bool:
             return False
 
         # Reconstruct total excl from taxes
-        total_vat = 0
-        reconstructed_total = 0
+        total_vat = 0.0
+        reconstructed_total = 0.0
         for tax in self.taxes:
             if tax.value is None or tax.rate is None or tax.rate == 0:
                 return False
@@ -254,7 +256,7 @@ def __taxes_plus_total_excl_match_total_incl(self) -> bool:
             return False
 
         # Reconstruct total_incl
-        total_vat = 0
+        total_vat = 0.0
         for tax in self.taxes:
             if tax.value is not None:
                 total_vat += tax.value
@@ -288,13 +290,11 @@ def __reconstruct_total_incl_from_taxes_plus_excl(self) -> None:
             multiplied by total_excl confidence
         """
         # Check total_tax, total excl exist and total incl is not set
-        if (
+        if not (
             self.total_excl.value is None
             or len(self.taxes) == 0
             or self.total_incl.value is not None
         ):
-            pass
-        else:
             total_incl = {
                 "value": sum(
                     [tax.value if tax.value is not None else 0 for tax in self.taxes]
@@ -319,17 +319,15 @@ def __reconstruct_total_excl_from_tcc_and_taxes(self) -> None:
             or len(self.taxes) == 0
             or self.total_excl.value is not None
         ):
-            pass
-        else:
-            total_excl = {
-                "value": self.total_incl.value
-                - sum(
-                    [tax.value if tax.value is not None else 0 for tax in self.taxes]
-                ),
-                "confidence": field_array_confidence(self.taxes)
-                * self.total_incl.confidence,
-            }
-            self.total_excl = Amount(total_excl, value_key="value", reconstructed=True)
+            return
+
+        total_excl = {
+            "value": self.total_incl.value
+            - sum([tax.value if tax.value is not None else 0 for tax in self.taxes]),
+            "confidence": field_array_confidence(self.taxes)
+            * self.total_incl.confidence,
+        }
+        self.total_excl = Amount(total_excl, value_key="value", reconstructed=True)
 
     def __reconstruct_total_tax_from_tax_lines(self) -> None:
         """
@@ -362,14 +360,11 @@ def __reconstruct_total_tax_from_incl_and_excl(self) -> None:
             or self.total_excl.value is None
             or self.total_incl.value is None
         ):
-            pass
-        else:
+            return
 
-            total_tax = {
-                "value": self.total_incl.value - self.total_excl.value,
-                "confidence": self.total_incl.confidence * self.total_excl.confidence,
-            }
-            if total_tax["value"] >= 0:
-                self.total_tax = Amount(
-                    total_tax, value_key="value", reconstructed=True
-                )
+        total_tax = {
+            "value": self.total_incl.value - self.total_excl.value,
+            "confidence": self.total_incl.confidence * self.total_excl.confidence,
+        }
+        if total_tax["value"] >= 0:
+            self.total_tax = Amount(total_tax, value_key="value", reconstructed=True)

mindee/documents/receipt.py

@@ -54,7 +54,7 @@ def __init__(
         )
 
     def __str__(self) -> str:
-        taxes = ", ".join(f"{t}" for t in self.taxes)
+        taxes = "\n       ".join(f"{t}" for t in self.taxes)
         return (
             "-----Receipt data-----\n"
             f"Filename: {self.filename}\n"
@@ -130,8 +130,8 @@ def __taxes_match_total(self) -> bool:
             return False
 
         # Reconstruct total_incl from taxes
-        total_vat = 0
-        reconstructed_total = 0
+        total_vat = 0.0
+        reconstructed_total = 0.0
         for tax in self.taxes:
             if tax.value is None or tax.rate is None or tax.rate == 0:
                 return False

mindee/fields/amount.py

@@ -27,7 +27,6 @@ def __init__(
             reconstructed=reconstructed,
             page_n=page_n,
         )
-
         try:
             self.value = round(float(amount_prediction[value_key]), 3)
         except (ValueError, TypeError, KeyError):

mindee/fields/base.py

@@ -66,7 +66,7 @@ def __eq__(self, other: Any) -> bool:
         return self.value == other.value
 
     def __str__(self) -> str:
-        if self.value:
+        if self.value is not None:
             return f"{self.value}"
         return ""
 

mindee/fields/tax.py

@@ -4,6 +4,7 @@
 
 
 class Tax(Field):
+    value: Optional[float]
     rate: Optional[float]
     code: Optional[str]
 

mindee/response.py

@@ -54,12 +54,12 @@ def format_response(
     logger.debug("Handling API response")
 
     # Create page level objects
-    for page_prediction in http_response["document"]["inference"]["pages"]:
+    for api_page in http_response["document"]["inference"]["pages"]:
         pages.append(
             doc_config.constructor(
-                api_prediction=page_prediction["prediction"],
+                api_prediction=api_page["prediction"],
                 input_file=input_file,
-                page_n=page_prediction["id"],
+                page_n=api_page["id"],
                 document_type=doc_config.document_type,
             )
         )

tests/documents/test_invoice.py

@@ -45,6 +45,7 @@ def test_all_na(invoice_object_all_na):
     assert invoice_object_all_na.locale.value is None
     assert invoice_object_all_na.total_incl.value is None
     assert invoice_object_all_na.total_excl.value is None
+    assert invoice_object_all_na.total_tax.value is None
     assert invoice_object_all_na.invoice_date.value is None
     assert invoice_object_all_na.invoice_number.value is None
     assert invoice_object_all_na.due_date.value is None